Feedback control undistorts body representations: A possible solution to the hand paradox

One of the most consistent findings in the study of body representation is that they are geometrically distorted. This finding is in contrast with human’s dexterity and fine motor skills, which are thought to be computationally optimal. How do we achieve optimal motor control given distorted body representations? This question, framed as the hand paradox, is still unresolved. While several solutions have been proposed at a conceptual level, none have addressed the paradox in the context of the actual computations performed by the sensorimotor system during motor control. In the present article, we propose that the hand paradox is solved by the closed loop nature of motor control, where body state estimates are an optimal combination of sensory information and internal predictions. We first formalize a dynamic Bayesian model of body representation during optimal feedback control. We then perform several simulations using this model to show how representations of finger geometry, which are initially distorted, become accurate very quickly after movement onset. This undistortion occurs regardless of initial levels of distortion and is incomplete in only the most unlikely conditions. These model simulations suggest that the closed-loop nature of sensorimotor control may be the key to resolve the hand paradox.